Metraipolis › archive › dc_4cfaefbb0219

Small Frequency Corrections Can Change What Survives KV Cache Compression — arXiv:2608.27128

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.27128
License (as archived): CC0-1.0
Archived: 2026-10-17T15:20:23.408Z
SHA-256 of the archived text: d3ebcdca127974e286c637ef501d1a3130bb35c4d21ce3ec25e6fea603138684
Stable URL of this page: https://metraipolis.com/documento/dc_4cfaefbb0219

How to cite

Small Frequency Corrections Can Change What Survives KV Cache Compression — arXiv:2608.27128 — arXiv (en), https://arxiv.org/abs/2608.27128 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4cfaefbb0219 (Metraipolis archive dc_4cfaefbb0219, sha256 d3ebcdca127974e286c637ef501d1a3130bb35c4d21ce3ec25e6fea603138684)

BibTeX

@misc{dc4cfaefbb0219,
  title = {Small Frequency Corrections Can Change What Survives KV Cache Compression — arXiv:2608.27128},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.27128},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4cfaefbb0219; sha256 d3ebcdca127974e286c637ef501d1a3130bb35c4d21ce3ec25e6fea603138684},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.27128. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Compressing a key-value cache before its next question is known requires choosing what to retain without knowing which evidence will matter. Value energy measures entry strength but does not distinguish isolated keys from those with many similar neighbors. We introduce TwinKV, a training-free method that discounts value energy by nonlocal post-RoPE key frequency. Prefix attention allocates head capacities, while retained entries preserve their original keys and values under an exact storage budget. Across four language models, TwinKV exceeds five evaluated compressed baselines in mean score on LongBench, LooGLE, and RULER at 50\% KV removal. Component controls isolate the frequency contribution. On Llama-3.2-1B RULER at 75\% removal, normalized frequency weights average 0.95, yet change 7\% of nonprotected retained positions and improve value-only retention by about 5.5 points under both uniform and adapted capacities. Permuting the weights within heads weakens this gain. These results show that modest frequency corrections can change retention and answering outcomes, with effects that depend on the model and task.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.27128 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T15:15:18.087Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).